Widemem Ai

remete61849stars

Local-first memory layer for LLM apps and agents. Runs offline on Ollama, sentence-transformers and FAISS; OpenAI and Anthropic optional. Importance scoring, contradiction resolution, confidence-aware

Knowledge Memory

Install Run on PlugMCP

GitHub Website

Run on PlugMCP

1# Widemem Ai

README.mdMCP server

Transport: stdio
Status: Local

Power your agent

Add this server

Install from the publisher. Not a PlugMCP endpoint.

Run on PlugMCP

Host a copy with our keys and logs.

No packaged release in the registry yet. Follow the repository README.

RepositoryPublisher installStep-by-step: add this in Claude

Remote URL or a command on your machine? How the two installs differ

Run on PlugMCP

Host a copy with our keys and logs.

A new server you create. It is not this listing's official endpoint. Use the publisher install when you want their URL and their credentials.

Launch

Server facts

Local

Transport: stdio
License: Apache-2.0
Last checked: 2026-10-28

Stars: 49
Tools: —
Uptime: —

Overview

About Widemem Ai

Tools

No tool list yet. Remote servers get one after the first health check.

FAQ

Questions

How do I install Widemem Ai?

Copy the stdio config for Cursor, Claude Desktop, Claude Code, or VS Code on this page. It uses the publisher's install, not a PlugMCP proxy.

Does Widemem Ai need an API key?

No required environment variable is listed on this record.

What transport does Widemem Ai use?

Widemem Ai runs locally over stdio.

When was Widemem Ai last checked?

Last checked 2026-10-28. Local-first memory layer for LLM apps and agents. Runs offline on Ollama, sentence-transformers and FAISS; OpenAI and Anthropic optional. Importance scoring, contradiction resolution, confidence-aware retrieval.

From the publisher

Documentation

widemem

Goldfish memory? ¬_¬ Fixed.

widemem is a local-first memory layer for LLM apps. It extracts facts from conversations, ranks them by importance and recency, resolves contradictions in one LLM call, and tells your agent how confident it is before it answers. By default everything runs on your machine: Ollama for the LLM, sentence-transformers for embeddings, FAISS and SQLite for storage. Cloud providers (OpenAI, Anthropic) are optional add-ons.

Install

pip install "widemem-ai[local]"   # FAISS, Ollama client, sentence-transformers
ollama pull llama3.1:8b           # the default local model, 4.9 GB

You need Ollama installed and running. The first run also downloads the all-MiniLM-L6-v2 embedding model (about 90 MB) from Hugging Face; after that, set HF_HUB_OFFLINE=1 and widemem runs fully offline. Python 3.10+.

What local costs: [local] pulls PyTorch through sentence-transformers (about 550 MB on macOS; more on Linux with CUDA wheels), and with the defaults each add() makes up to two LLM calls, extract then resolve. On an Apple M4 with 32 GB, an add() took 5 to 17 seconds and a search 0.04 seconds. The default is llama3.1:8b because llama3.2 (3B) kept the stale fact and split "I'm allergic to penicillin" into one fact per word in every one of our test runs.

Upgrading from 1.6 or earlier: the defaults moved from OpenAI to local, and a 1536-dimension FAISS index will not load under the 384-dimension default embedder. Set provider="openai" on both LLMConfig and EmbeddingConfig to keep the old behavior. Details in the changelog.

Quick start

from widemem import WideMemory, MemoryConfig
from widemem.core.types import VectorStoreConfig

memory = WideMemory(MemoryConfig(
    vector_store=VectorStoreConfig(provider="faiss", path="./widemem_data"),
))

memory.add("I live in San Francisco and work as a software engineer", user_id="alice")
memory.add("I just moved to Boston", user_id="alice")   # checked against stored facts: ADD, UPDATE or DELETE

results = memory.search("where does alice live", user_id="alice")
if results:
    print(results[0].memory.content)                   # results.confidence says how sure it is
    history = memory.get_history(results[0].memory.id) # every write is logged

Without a path, FAISS keeps vectors in RAM and they are gone when the process exits; the SQLite history persists either way. WideMemory also works as a context manager.

Cloud providers, if you want them

No memory, prompt or embedding leaves your machine unless you configure a cloud provider. Each provider picks its own default model.

pip install "widemem-ai[openai,faiss]"      # or [anthropic,local] for Claude + local embeddings

from widemem import WideMemory, MemoryConfig
from widemem.core.types import EmbeddingConfig, LLMConfig

# OpenAI for both: gpt-4o-mini and text-embedding-3-small
memory = WideMemory(MemoryConfig(
    llm=LLMConfig(provider="openai"),
    embedding=EmbeddingConfig(provider="openai"),
))

# Claude for extraction; embeddings and vectors stay local
memory = WideMemory(MemoryConfig(llm=LLMConfig(provider="anthropic")))

Set OPENAI_API_KEY or ANTHROPIC_API_KEY. Mixing is fine: a cloud LLM with local embeddings keeps every vector on your machine.

Layer	Default (local)	Options
LLM	Ollama `llama3.1:8b`	OpenAI, Anthropic, any Ollama model
Embeddings	sentence-transformers `all-MiniLM-L6-v2`	Ollama `nomic-embed-text`, OpenAI
Vectors	FAISS	Qdrant (embedded, `localhost:6333` or a remote `url`), pgvector
History	SQLite

Every field and default: docs/configuration.md.

How widemem differs

Feature	What it does
Importance and decay	The LLM rates each fact 1 to 10. Search blends similarity, importance and recency, with weights adapted to the question type.
Batch conflict resolution	New facts and the related stored memories go to the LLM in one call, which decides ADD, UPDATE, DELETE or no-op for each.
YMYL protection	Opt-in. Health, financial, legal and safety facts get an importance floor of 8, no decay, and forced contradiction checks. Regex catches the obvious cases; the extraction LLM tags the implied ones.
Confidence	Every search returns HIGH, MODERATE, LOW or NONE, so an agent can say "I don't have that" instead of guessing.
Hierarchy	`summarize()` rolls facts into summaries and themes (a no-op under 10 facts unless `force=True`). In `balanced` and `deep` mode, broad questions then get themes and specific ones get facts.
Retrieval modes	`fast`, `balanced` and `deep` retrieve 10, 25 or 50 memories.
Audit log	Adds, updates, deletes, imports and pins are logged to SQLite with the content on both sides. Entries record what changed and when, not who.

How each one works, with snippets: docs/guide.md. YMYL details and limits: YMYL.md.

Benchmark

On LoCoMo's 1,540 answerable questions (the adversarial category excluded, as for every system on the comparison page), widemem 1.5.0 scored 55.15% with an independent GPT-4o judge (56.32% self-graded), using about 213 tokens of context per query. That run used OpenAI (gpt-4o-mini, text-embedding-3-small) with top_k=10 per speaker; the local default has not been benchmarked yet. It places widemem in the lower half of the eight systems on widemem.ai/benchmarks, at a small fraction of the context most of them use.

Per-category labels published before 2026-07-06 were wrong; the correction is in the corrections log.

The runner (benchmark/run_ws1.py) and question split are in this repo. To re-run it, clone snap-research/locomo into benchmark/locomo-data/ (the runner reads data/locomo10.json) and set OPENAI_API_KEY.

Integrations

MCP server for Claude Desktop, Cursor and other MCP clients. Runs on the local stack by default.

pip install "widemem-ai[mcp,local]"
python -m widemem.mcp_server

Tools exposed: widemem_add, widemem_search, widemem_delete, widemem_count, widemem_pin, widemem_export, widemem_health. Setup and environment variables: docs/mcp.md.

LangChain. widemem.integrations.langchain.WidememRetriever is a BaseRetriever. Install [langchain]; example in examples/langchain_retriever.py.

REST server. pip install "widemem-ai[server,local]", then python -m widemem.server. It refuses to start when WIDEMEM_HOST is non-local and WIDEMEM_API_KEY is unset. The check reads WIDEMEM_HOST, not the bind address, so set it even when you launch uvicorn yourself. In Docker, the image talks to Ollama for both the LLM and embeddings; point them at Ollama on the host with WIDEMEM_LLM_BASE_URL and WIDEMEM_EMBEDDING_BASE_URL (http://host.docker.internal:11434). (Linux Docker Engine: also pass --add-host=host.docker.internal:host-gateway)

API

Method	Does
`add(text, user_id, ...)` / `add_batch(texts, ...)`	Extract, resolve and store facts
`search(query, user_id, top_k, mode, explain, ...)`	Ranked results with `.confidence`; `explain=True` returns a scored breakdown
`search_stream(...)`	Async generator, approximate order
`pin(text, user_id, importance=9.0)`	Store a fact the user flagged as important
`get`, `delete`, `count`, `get_history`	The usual
`purge_expired(older_than_days, ...)`	Delete old memories for good (YMYL kept unless asked)
`export_json` / `import_json`	Move memories between stores
`summarize(user_id, force)`	Build summaries and themes

Full signatures: docs/api.md.

Scope and safety

widemem is developer infrastructure, provided under Apache 2.0 as is. It is not medical, legal, tax or financial advice, and not a medical device. YMYL handling is a best-effort heuristic, the prompt-injection sanitizer is a baseline defense on extraction input, and the history log does not record callers. Keep a human in the loop for high-stakes decisions. When you self-host, your data stays in your environment and we receive nothing. The software may be subject to export-control and sanctions laws (including the US EAR and OFAC lists).

Roadmap

Open issues; vote with reactions:

#21 Source-message provenance: link each fact in the history log to the message that produced it
#22 LangChain BaseChatMessageHistory adapter
#24 LangGraph BaseStore adapter

Managed hosting is available on request: [email protected].

Not planned: more vector backends beyond FAISS, Qdrant and pgvector; a self-serve multi-tenant service; a web UI; a GraphQL API; a memory-management CLI.

Development

git clone https://github.com/remete618/widemem-ai
cd widemem-ai
pip install -e ".[all,dev]"
pytest

Contact and license

[email protected] · widemem.ai · issues

Apache 2.0, see LICENSE. Website terms: widemem.ai/terms.

Widemem Ai

remete61849stars

Knowledge Memory

Install Run on PlugMCP

GitHub Website

Run on PlugMCP

1# Widemem Ai

README.mdMCP server

Transport: stdio
Status: Local

Power your agent

Add this server

Install from the publisher. Not a PlugMCP endpoint.

Run on PlugMCP

Host a copy with our keys and logs.

No packaged release in the registry yet. Follow the repository README.

RepositoryPublisher installStep-by-step: add this in Claude

Remote URL or a command on your machine? How the two installs differ

Run on PlugMCP

Host a copy with our keys and logs.

A new server you create. It is not this listing's official endpoint. Use the publisher install when you want their URL and their credentials.

Launch

Server facts

Local

Transport: stdio
License: Apache-2.0
Last checked: 2026-10-28

Stars: 49
Tools: —
Uptime: —

Overview

About Widemem Ai

Tools

No tool list yet. Remote servers get one after the first health check.

FAQ

Questions

How do I install Widemem Ai?

Copy the stdio config for Cursor, Claude Desktop, Claude Code, or VS Code on this page. It uses the publisher's install, not a PlugMCP proxy.

Does Widemem Ai need an API key?

No required environment variable is listed on this record.

What transport does Widemem Ai use?

Widemem Ai runs locally over stdio.

When was Widemem Ai last checked?

From the publisher

Documentation

widemem

Goldfish memory? ¬_¬ Fixed.

Install

pip install "widemem-ai[local]"   # FAISS, Ollama client, sentence-transformers
ollama pull llama3.1:8b           # the default local model, 4.9 GB

Quick start

from widemem import WideMemory, MemoryConfig
from widemem.core.types import VectorStoreConfig

memory = WideMemory(MemoryConfig(
    vector_store=VectorStoreConfig(provider="faiss", path="./widemem_data"),
))

memory.add("I live in San Francisco and work as a software engineer", user_id="alice")
memory.add("I just moved to Boston", user_id="alice")   # checked against stored facts: ADD, UPDATE or DELETE

results = memory.search("where does alice live", user_id="alice")
if results:
    print(results[0].memory.content)                   # results.confidence says how sure it is
    history = memory.get_history(results[0].memory.id) # every write is logged

Without a path, FAISS keeps vectors in RAM and they are gone when the process exits; the SQLite history persists either way. WideMemory also works as a context manager.

Cloud providers, if you want them

No memory, prompt or embedding leaves your machine unless you configure a cloud provider. Each provider picks its own default model.

pip install "widemem-ai[openai,faiss]"      # or [anthropic,local] for Claude + local embeddings

from widemem import WideMemory, MemoryConfig
from widemem.core.types import EmbeddingConfig, LLMConfig

# OpenAI for both: gpt-4o-mini and text-embedding-3-small
memory = WideMemory(MemoryConfig(
    llm=LLMConfig(provider="openai"),
    embedding=EmbeddingConfig(provider="openai"),
))

# Claude for extraction; embeddings and vectors stay local
memory = WideMemory(MemoryConfig(llm=LLMConfig(provider="anthropic")))

Set OPENAI_API_KEY or ANTHROPIC_API_KEY. Mixing is fine: a cloud LLM with local embeddings keeps every vector on your machine.

Layer	Default (local)	Options
LLM	Ollama `llama3.1:8b`	OpenAI, Anthropic, any Ollama model
Embeddings	sentence-transformers `all-MiniLM-L6-v2`	Ollama `nomic-embed-text`, OpenAI
Vectors	FAISS	Qdrant (embedded, `localhost:6333` or a remote `url`), pgvector
History	SQLite

Every field and default: docs/configuration.md.

How widemem differs

Feature	What it does
Importance and decay	The LLM rates each fact 1 to 10. Search blends similarity, importance and recency, with weights adapted to the question type.
Batch conflict resolution	New facts and the related stored memories go to the LLM in one call, which decides ADD, UPDATE, DELETE or no-op for each.
YMYL protection	Opt-in. Health, financial, legal and safety facts get an importance floor of 8, no decay, and forced contradiction checks. Regex catches the obvious cases; the extraction LLM tags the implied ones.
Confidence	Every search returns HIGH, MODERATE, LOW or NONE, so an agent can say "I don't have that" instead of guessing.
Hierarchy	`summarize()` rolls facts into summaries and themes (a no-op under 10 facts unless `force=True`). In `balanced` and `deep` mode, broad questions then get themes and specific ones get facts.
Retrieval modes	`fast`, `balanced` and `deep` retrieve 10, 25 or 50 memories.
Audit log	Adds, updates, deletes, imports and pins are logged to SQLite with the content on both sides. Entries record what changed and when, not who.

How each one works, with snippets: docs/guide.md. YMYL details and limits: YMYL.md.

Benchmark

Per-category labels published before 2026-07-06 were wrong; the correction is in the corrections log.

Integrations

MCP server for Claude Desktop, Cursor and other MCP clients. Runs on the local stack by default.

pip install "widemem-ai[mcp,local]"
python -m widemem.mcp_server

Tools exposed: widemem_add, widemem_search, widemem_delete, widemem_count, widemem_pin, widemem_export, widemem_health. Setup and environment variables: docs/mcp.md.

LangChain. widemem.integrations.langchain.WidememRetriever is a BaseRetriever. Install [langchain]; example in examples/langchain_retriever.py.

API

Method	Does
`add(text, user_id, ...)` / `add_batch(texts, ...)`	Extract, resolve and store facts
`search(query, user_id, top_k, mode, explain, ...)`	Ranked results with `.confidence`; `explain=True` returns a scored breakdown
`search_stream(...)`	Async generator, approximate order
`pin(text, user_id, importance=9.0)`	Store a fact the user flagged as important
`get`, `delete`, `count`, `get_history`	The usual
`purge_expired(older_than_days, ...)`	Delete old memories for good (YMYL kept unless asked)
`export_json` / `import_json`	Move memories between stores
`summarize(user_id, force)`	Build summaries and themes

Full signatures: docs/api.md.

Scope and safety

Roadmap

Open issues; vote with reactions:

#21 Source-message provenance: link each fact in the history log to the message that produced it
#22 LangChain BaseChatMessageHistory adapter
#24 LangGraph BaseStore adapter

Managed hosting is available on request: [email protected].

Not planned: more vector backends beyond FAISS, Qdrant and pgvector; a self-serve multi-tenant service; a web UI; a GraphQL API; a memory-management CLI.

Development

git clone https://github.com/remete618/widemem-ai
cd widemem-ai
pip install -e ".[all,dev]"
pytest

Contact and license

[email protected] · widemem.ai · issues

Apache 2.0, see LICENSE. Website terms: widemem.ai/terms.

Add this server

Host a copy with our keys and logs.

Server facts

About Widemem Ai

Tools

Questions

How do I install Widemem Ai?

Does Widemem Ai need an API key?

What transport does Widemem Ai use?

When was Widemem Ai last checked?

Documentation

widemem

Install

Quick start

Cloud providers, if you want them

How widemem differs

Benchmark

Integrations

API

Scope and safety

Roadmap

Development

Further reading

Contact and license

Add this server

Host a copy with our keys and logs.

Server facts

About Widemem Ai

Tools

Questions

How do I install Widemem Ai?

Does Widemem Ai need an API key?

What transport does Widemem Ai use?

When was Widemem Ai last checked?

Documentation

widemem

Install

Quick start

Cloud providers, if you want them

How widemem differs

Benchmark

Integrations

API

Scope and safety

Roadmap

Development

Further reading

Contact and license